Our blog

Tweetminster Predicts - findings

Tweetminster Predicts: Findings

Posted at Wed, May 12th 2010, 07:00
permalink

Tweetminster Predicts - a hung parliament

Today we’re releasing the final set of ‘Tweetminster Predicts’ figures, our experiment in predictive modelling that studies the correlation between buzz on Twitter and election results. The model looks at the most mentioned candidate in each of the constituencies represented on Twitter (full details of the methodology are below).

Our latest figures, based on the 433 constituencies now mentioned on Twitter, don’t show significant changes since last week, with topline figures - CON 35% (nc) LAB 30% (nc) LDEM 27% (+1).
With a UNS, these figures would point to a hung parliament with Labour 45 seats short of a majority. 

When we launched our study last month, we outlined several trends in the data, including:

• The Liberal Democrats performing better than polls were showing (they were below 20% then), especially strong in the South West in LibCon marginals
• A decline in SNP support when compared to recent polls and forecasts
• No significant shifts towards Plaid Cymru
• A strong Conservative performance in the East Midlands. 

In addition to confirming these trends, the latest set of data also points to:

• Growing support for the Conservatives in Wales
• The high likelihood of Caroline Lucas winning in Brighton Pavilion
• Esther Rantzen winning in Luton South
• While as expected several Labour seats will swing to the Conservatives and the Liberal Democrats, the data suggests that a few will swing the other way too. 

This week we are also releasing predictions for 266 constituencies around which significant buzz has been generated over the past month:

We have also updated the paper of the study:

Tweetminster Predicts

Following the election we will publish the complete findings of the study. The final set of predictions will also be uploaded to our site shortly.

Posted at Wed, May 5th 2010, 08:00
permalink

Help us map voter turnout on May 6th

On May 6th millions of people across the county will head to the polling booths to cast their vote. Tens of thousands of them use Twitter. Tweetminster, Channel 4, The Guardian and the New Statesman are teaming up to map the day’s action and we need your help.

On May 6th, when you head out to vote, tweet #ukvote alongside the first half of your post code (for example #ukvote SE1) and we will plot votes in real time on a map. 

We’re not asking you to include who you’re voting for - that’s a secret you don’t need to share. 

The map and data will be made available online and anyone will be able to easily add it to their site. We will also showcase a regional breakdown of turnout.

The goal of the experiment is to inspire more people to vote and to help get a sense of turnout (on Twitter at least) during the course of the day and across the country.

The experiment is inspired by a similar initiative around the hashtag #uksnow. 

We’ll tweet a reminder ahead of the big day. In the mean time, thanks for your help!

Posted at Mon, May 3rd 2010, 13:10
permalink

Leaders' Debate Vol. 3

The third, and final, debate saw an increase in active compared to the previous debate. While the total number of tweets & tweeters didn’t reach the volume of the opening debate, the debate though registered highest peak, with 51.7 tweets per second posted as the leaders debated immigration. 

• Total number of tweets: 154,342 (+11,546 compared to the second debate)
• Average frequency of tweets: 26.77 tweets per second (+0.1)
• Number of tweeters: 33,095 (+4,305)
• Peak: 51.7 tweets per second (+18.52)

Sentiment scores: Clegg 3.13 (-0.5 change from the beginning of the debate), Brown 2.99 (-0.15), Cameron 2.96 (-0.22). Once again, Brown and Cameron scored very closely. While Clegg scored higher, his score dropped the most during the debate, possibly meaning that his performance didn’t match very high expectations.

Update: The morning after the debate sees Cameron and Clegg on 3.26 and Brown on 2.86.

The most tweeted issues:

• Immigration
• Tax
• Banks
• Jobs
• Euro
• Education
• Amnesty
• Cuts
• Manufacturing
• Benefits
• Housing
• Bonus

The following chart shows the frequency of tweets per second over the course of the debate:

Analysis of the first debate can be found here While that of the second debate is here

Posted at Thu, Apr 29th 2010, 22:50
permalink